Key points
- Revenue is visitors times conversion rate times average selling price. A fall in any one of them lowers sales, and the drop can be split into three amounts of money that add up exactly to the total. [1]
- Start with whichever amount is largest. Buying ads while conversion is falling only buys more visitors who do not order.
- Before acting, confirm both periods are measured the same way: platforms redefine metrics, and percentages on low-traffic products swing by chance. [3]
Revenue has only three multipliers
Whatever platform you sell on, a product's revenue can be written the same way: visitors × conversion rate × average selling price. Amazon's Business Reports call them Sessions, Units Ordered and Ordered Product Sales [1]; TikTok Shop's product analytics [4], Shopee's Business Insights and the analytics on your own site use other names for the same structure.
The formula is useful because it rules things out. When sales fall, teams tend to decide that traffic is short and go straight to budgets and creators. But a decline in any of the three produces the same symptom. Knowing which one moved is what gives the next step a reason.
Split the drop into three amounts
Seller Labs describes a split you can apply directly: break the change between two periods into a traffic effect, a conversion effect and a price effect, each expressed in money. [1]
| Effect | How to calculate it | Result in the example |
|---|---|---|
| Traffic | Change in visitors × previous conversion rate × previous price | +$720 |
| Conversion | Current visitors × change in conversion rate × previous price | −$3,150 |
| Price | Current units × change in price | $0 |
The example comes from the same article. A $30 product had 4,000 visitors in August at a 12.0% conversion rate: 480 units and $14,400. In September it had 4,200 visitors at 9.5%: 399 units and $11,970, down $2,430. Split apart, traffic actually added $720; the entire drop came from conversion. [1]
Spending another $1,000 on ads at $1.25 a click would buy only 800 clicks, whereas restoring the 12.0% conversion rate on the same visitors is worth $3,150. [1] The two directions pay back very differently, which is the reason to do the split.
Investigate the largest effect first
Once the amounts are known, the numbers set the order of investigation, not habit. The table groups the usual destinations by largest effect. [1]
| Largest effect | Check first | Common causes |
|---|---|---|
| Traffic is negative | Ad budget and delivery status, organic search rank, the posting rhythm of content and creators | Budget ran out, rank slipped, a video or live stream that drove visits stopped |
| Conversion is negative, stock and the buy button are normal | The product page, recent reviews, the price gap to competitors | The main image or title changed, a bad review appeared, a competitor cut prices |
| Conversion is negative and stock or the default buying position was lost | Inventory, available variants, your offer | The main variant sold out, another seller took the buying position |
| Conversion is negative while ad traffic is growing | Which keywords and audiences the new traffic came from | Scaling up brought in visitors with weaker intent |
| Price is negative | Promotions, vouchers and stacked discounts | The price was not restored after a campaign, or discounts stacked further than planned |
Add-to-carts but no orders: the problem is at checkout
A conversion drop divides again into two cases. Visitors who do not add to cart were not persuaded by the product page. Visitors who add to cart and do not order were lost between the cart and payment, and the two are investigated in entirely different places. [2]
Noibu cites Baymard Institute research showing that the top reason US shoppers abandon checkout is extra costs that are too high, at 40%; site errors or crashes account for 17%. [2] For marketplace sellers the matching checks are shipping fees, minimum-spend thresholds, whether a voucher is actually applying, and the estimated delivery date. Sellers with their own site should also compare drop-off at each step by device and browser and start with the steepest one.
First make sure it is the same ruler
Some drops never happened; only the measurement changed. These are worth checking before any analysis.
- Platforms change metric definitions. According to BigSeller's summary of Shopee's Seller Education Hub, Shopee Malaysia changed Sales in January 2026 to the amount after seller rebates, still including cancelled and refunded orders. Comparing old and new figures directly shows a fall that is not there. [3]
- Do not average daily conversion rates. Add up units and visitors for the whole period, then divide. [1]
- Be careful with percentages on low-traffic products. At 300 visitors a one-point change in conversion is only three units, which is ordinary noise; use a 60- or 90-day window for these. [1]
- For seasonal categories, the same month last year is a more reliable comparison than last month. [1]
The order of a review
Put together, a full check runs in this order. Each step yields a definite conclusion before you decide whether to go on.
- Confirm both periods cover the same number of days, the same time zone and the same metric definitions, and mark the dates of campaigns, price changes and stock-outs.
- Rank products by the amount of revenue lost and find the few that account for most of the drop.
- For each of them, split the change into traffic, conversion and price.
- Start with the largest amount, change one variable at a time, and record the date of the change and the date of the next check.
- At that check, run the same split again and confirm the effect has narrowed.
The method needs no particular tool; a spreadsheet is enough. To save the exporting and matching, you can also hand both periods to Aurovia Agent and ask it to split the three effects product by product and label where each figure came from.
Sources
This article draws on the public material below, reorganised for marketplace sellers. For the names and definitions of metrics, the seller centre you actually use is the authority.
- Seller Labs: Sales Dropped? Split It Before You Raise Ad Spend
- Noibu: Traffic and Add-to-Carts but Few Orders? Find Where Checkout Breaks
- BigSeller: Shopee Malaysia Updates Sales Data Reporting Across Seller Tools
- TikTok Shop Seller Academy: using product analytics (US)
Sources checked on
Ongoing tasksSample data